This thesis details the development and testing of a new scale for measuring human attachment to dogs which allows for the measurement of weaker attachment levels as well as stronger ones (the CDA scale). The correlation between dog-owner’s scores on the CDA scale and their dog’s actual attachment behaviour is assessed and discussed, as well as the dog-owners limited ability to predict the behaviour of their dog in a controlled situation (the Strange Situation Test (SST)) whereby the dogs meet a previously unknown person. The CDA scale was formed by utilising items from pre-existing scales (the Comfort from Companion Animals scale and the Lexington Attachment to Pets scale), trialed on the internet with a large self-selected sample of dog-owners and analysed and reduced using factor analysis. The CDA was completed with the addition of some negative items derived from a small sample of dog-owners who expressed drawbacks to keeping a dog. In addition 100 people living with dogs they did not consider themselves to be the primary carer of, and 100 people with dogs they considered to have behavioural problems also completed the CDA to allow for the assessment of reliability and validity, and for consideration of the possible links between human perceptions of attachment/dog behavioural problems and actual scores on the CDA. Dogs’ attachment behaviour was assessed by cluster analysis of behaviours observed in the SSD: 51 dog-owner pairs took part in the study which revealed a number of secure-base behavioural categories analogous to those typically observed in human mother-infant interactions in Ainsworth’s original (1969) SST. In addition five captive wolves were also observed in a modified version of the SST. Data from these observations is discussed in a case-wise manner and it is clear that captive wolves do not exhibit the suite of attachment behaviours (to their familiar handler) as previously observed in the dog study. However, the wolves’ familiar handler was very adept at predicting the behaviour of his wolves in this situation. These findings are important in furthering our understanding of human-canine attachments in general, but especially given the number of dog-owner pairings which appear to fail due to poor or misunderstood attachments. An effective attachment scale for people, and a valid measure and analysis of attachment behaviour in dogs is a further development in ensuring successful pairings of people with dogs in a variety of contexts such as pet dogs and service dogs.